10 Tools
Query, search, write, patch, and list vault content. Delegate tasks to cheaper models. Track health and budget.
Every session starts from scratch.
Your AI assistant doesn’t remember that you chose PostgreSQL over MySQL last month. Or that your deploy pipeline has a quirk with Redis cache. You end up repeating yourself, pasting the same context, stuffing everything into a giant CLAUDE.md that eats thousands of tokens before you even ask a question.
Hive fixes this.
You keep a folder of Markdown files — project decisions, architecture notes, lessons learned, task lists. Hive lets your assistant query them on demand. Only what’s relevant, when it’s relevant.
But the real power isn’t per-project memory.
It’s the shared practices. You define your coding standards, architecture patterns, and review checklists once. Every project — personal side projects, your day job, your team’s repos — follows the same playbook. That consistency compounds. And it’s what actually makes you faster.
One more thing: this doesn’t have to be an Obsidian vault. Any folder with .md files works. Your knowledge lives in plain files you control, not locked inside a tool.
10 Tools
Query, search, write, patch, and list vault content. Delegate tasks to cheaper models. Track health and budget.
5 Resources
MCP resources for projects, health metrics, context, tasks, and lessons — directly consumable by AI clients.
4 Prompts
Built-in protocols for retrospectives, delegation, vault sync, and benchmarking.
Worker Routing
Automatic tier selection: local Ollama, OpenRouter free, OpenRouter paid, or reject. $1/mo budget cap.
claude mcp add -s user hive -- uvx --upgrade hive-vaultAdd to ~/.codex/config.toml:
[mcp_servers.hive-vault]command = "uvx"args = ["--upgrade", "hive-vault"]gemini mcp add -s user hive-vault uvx -- --upgrade hive-vaultAdd to .vscode/mcp.json:
{ "servers": { "hive-vault": { "command": "uvx", "args": ["--upgrade", "hive-vault"] } }}No cloning, no venv, no setup. Learn more →
The highest-value setup:
Your assistant writes to the vault → Obsidian Git auto-commits → next session, everything is there. No manual sync.
Hive works with any Markdown directory — Obsidian is recommended, not required.
max_lines calibration.